microsoft/mscclpp v0.8.0

MSCCL++ v0.8.0

Repository: microsoft/mscclpp

Tag: v0.8.0

Published: 2025-10-10T18:32:33Z

Prerelease: no

Release notes:

What's Changed

• Fix #458 by @Binyang2014 in https://github.com/microsoft/mscclpp/pull/568
• Fix multinode test failure by @Binyang2014 in https://github.com/microsoft/mscclpp/pull/574
• Separate linters from cmake by @chhwang in https://github.com/microsoft/mscclpp/pull/587
• Fix relaxedWait() by @chhwang in https://github.com/microsoft/mscclpp/pull/594
• NCCL fixes by @chhwang in https://github.com/microsoft/mscclpp/pull/592
• Updated Dev Container by @chhwang in https://github.com/microsoft/mscclpp/pull/591
• Support CudaIpc connection within a single process by @chhwang in https://github.com/microsoft/mscclpp/pull/593
• Fix GpuStreamPool to be aware of the device ID of streams by @chhwang in https://github.com/microsoft/mscclpp/pull/590
• update pytest and python API to fix ut failure by @Binyang2014 in https://github.com/microsoft/mscclpp/pull/598
• Fixed the local channel test by @chhwang in https://github.com/microsoft/mscclpp/pull/597
• Use smart pointer for IB structure by @Binyang2014 in https://github.com/microsoft/mscclpp/pull/585
• Update documentation by @chhwang in https://github.com/microsoft/mscclpp/pull/576
• Support CUDA 12.9 by @chhwang in https://github.com/microsoft/mscclpp/pull/600
• Merge ChannelTrigger with ProxyTrigger by @chhwang in https://github.com/microsoft/mscclpp/pull/601
• MNNVL fix by @chhwang in https://github.com/microsoft/mscclpp/pull/604
• New DSL implementation by @Binyang2014 in https://github.com/microsoft/mscclpp/pull/579
• python doc auto generation by @chhwang in https://github.com/microsoft/mscclpp/pull/605
• all2all implementation by @Binyang2014 in https://github.com/microsoft/mscclpp/pull/609
• Fix ut by @Binyang2014 in https://github.com/microsoft/mscclpp/pull/613
• Create ib mr for per ib transport by @Binyang2014 in https://github.com/microsoft/mscclpp/pull/611
• Fix for multi-nodes test by @Binyang2014 in https://github.com/microsoft/mscclpp/pull/614
• add torch test by @Binyang2014 in https://github.com/microsoft/mscclpp/pull/612
• AlltoAll Test Support by @caiomcbr in https://github.com/microsoft/mscclpp/pull/606
• Adding Channel Id Field DSL Port Channel Operations by @caiomcbr in https://github.com/microsoft/mscclpp/pull/615
• Fix deadlock in Executor channel setup by @caiomcbr in https://github.com/microsoft/mscclpp/pull/616
• Fix NVLS correctness issue by @Binyang2014 in https://github.com/microsoft/mscclpp/pull/618
• Fixed cpp linter by @chhwang in https://github.com/microsoft/mscclpp/pull/619
• Thread Block Group DSL by @caiomcbr in https://github.com/microsoft/mscclpp/pull/621
• Fix memory exchange within a single process by @chhwang in https://github.com/microsoft/mscclpp/pull/624
• Fix hang issue in logging submodule by @Binyang2014 in https://github.com/microsoft/mscclpp/pull/625
• Integrate MSCCL++ with torch workload by @Binyang2014 in https://github.com/microsoft/mscclpp/pull/626
• Add FifoDeviceHandle::poll() by @chhwang in https://github.com/microsoft/mscclpp/pull/630
• Fix Illegal Memory Access in nvls_test for CUDA12.9 by @abhijangda in https://github.com/microsoft/mscclpp/pull/631
• Adapt with torch 2.6 by @Binyang2014 in https://github.com/microsoft/mscclpp/pull/632
• Fix for safe process teardown by @chhwang in https://github.com/microsoft/mscclpp/pull/633
• use unix socket to share fd by @Binyang2014 in https://github.com/microsoft/mscclpp/pull/634
• Address teardown issue by @Binyang2014 in https://github.com/microsoft/mscclpp/pull/638
• Revise NCCL API implementation by @Binyang2014 in https://github.com/microsoft/mscclpp/pull/617
• Support detailed version tracking that captures git repository information by @seagater in https://github.com/microsoft/mscclpp/pull/639
• Fix Rocm build issue by @Binyang2014 in https://github.com/microsoft/mscclpp/pull/642
• Add 2 Node AllReduce DSL Algorithm by @caiomcbr in https://github.com/microsoft/mscclpp/pull/636
• Make ncclReduce/ncclSend/ncclRecv work by @Binyang2014 in https://github.com/microsoft/mscclpp/pull/643
• Reduce memory footprint for allreduce8 and allgather6 by @Binyang2014 in https://github.com/microsoft/mscclpp/pull/644
• Add MSCCLPP_GIT_COMMIT micro by @Binyang2014 in https://github.com/microsoft/mscclpp/pull/640
• Address corner case when generating version file by @Binyang2014 in https://github.com/microsoft/mscclpp/pull/641
• Pipeline fix by @Binyang2014 in https://github.com/microsoft/mscclpp/pull/645

New Contributors

• @abhijangda made their first contribution in https://github.com/microsoft/mscclpp/pull/631

Full Changelog: https://github.com/microsoft/mscclpp/compare/v0.7.0...v0.8.0